The state of document approvals: what 455 people told us

What we found

Approval has drifted out of the document. Email is where sign-off most often happens, on 49.9%, ahead of the document or design tool itself on 45.1%. The split is telling: among people whose work spans several platforms, email wins comfortably. Among the minority working in a single platform, most approve in the document. If the work sits in one place, people approve it where it sits.

Almost everyone has a problem with it. Nineteen people out of 455 said their approval process gives them no trouble. Everyone else named something, led by feedback scattered across tools and slow turnaround.

Nobody gets to work in just one tool. 61% said their approval-bound content lives in more than one platform. Of the people using Microsoft 365 for it, more than four in ten also use Google Drive, and half of teams run both suites at once.

Fragmentation lands on the people who can least afford it. People spread across platforms lose track of feedback far more often than people in a single tool, 63% against 42%, and they are markedly more likely to need a formal audit trail, 81% against 60%. Both gaps are statistically strong.

Accountability climbs with seniority. Half of individual contributors need formal approval records. For people managers and above it is closer to nine in ten. The person who gets asked for the audit trail is usually not the person who wrote the document.

Microsoft 365 is where people want it fixed. 41.5% chose it the first time we asked and 41.6% the second, from two differently worded questions twelve days apart. That is the most stable result in the study.

Why email cannot carry this

An approval needs someone with authority to say yes, to a specific version, on a date, in a way anyone can check later. An inbox gives you none of that. There is no field for “approved”, only a sentence that might read “looks good” or “fine by me”. Finding out where a document stands means scrolling back through a thread; getting it finished means leaving email for Teams or Slack to chase whoever has not replied. What got approved was an attachment, frozen when it was sent, while the live document moved on. And a thread is not an audit trail: it sits in private mailboxes and leaves when its owner does.

Set the five most common complaints in our survey against that list and they line up almost exactly. They are not five problems. They are one problem described five ways, which is that a decision about a document is being kept somewhere the document is not.

The alternative some teams take

If the problem is scattered work, one answer is to stop it being scattered. Move everyone onto a single platform and that platform’s approval feature covers the whole job. One customer we interviewed has spent a year doing exactly that, and told us she did not want a cross-platform view because it would undo her work.

That is a legitimate solution, with real benefits beyond approvals. It also has a cost: teams end up working in a tool they would not have chosen. And with half of organisations running two document suites at once, consolidation is a multi-year programme rather than a decision. The teams who manage it have solved this. Everyone else needs the approval to travel instead.

What tooling can and cannot fix

Two results argue against software solving this on its own. The 23 respondents who already use a dedicated approval or workflow tool reported the same problems at the same rates as everyone else, and none of them reported no problems at all. With 23 people that is not a statistically meaningful gap, so read it as no measurable improvement rather than as evidence those tools make things worse. The likely reason is structural: they each govern one platform.

The second is blunter. One customer, whose process has never been audited, told us their documents “are not really looked at with too much accuracy” and that people are “skipping through”. Where nobody checks the record, an approval trail is decoration.

But skim-approving is partly a tooling failure. People skim because they are handed a whole document and left to work out what changed and whether any of it concerns them. Showing a diff against the last approved version, summarising what moved, and dropping approvers whose area a revision does not touch all put attention where it belongs. Asking fewer people is faster than chasing more of them.

What needs to change

Approval state belongs with the document, so anyone opening the file can see whether it was signed off, by whom, and against which version. It has to span the tools an organisation actually uses, because the people carrying the most compliance risk are the ones spread across the most platforms. The record has to be exportable as evidence without anyone rebuilding it from an inbox. And it has to earn the approver’s attention rather than assume it.

That is what we are building. We started with Word and Confluence, and we are looking for design partners whose approvals look like the ones in this research.

About the research

We ran the same survey twice on Prolific in February 2026, on the 6th and the 18th, so we could see which answers held up. Both screened for people involved in a document or design approval in the previous three months and removed anyone who was not. 559 people responded and 455 qualified. Only the second run asked the demographic questions, so those figures describe 201 people and the report says so each time. Percentages are calculated against those who qualified and answered, never against everyone who responded.

The quotes come from a separate and much smaller study: five recorded interviews with existing customers of our Confluence approvals app. It is not connected to the survey and cannot show how common anything is. It is there to put words to what the survey counts.

Full method, limits and the question-by-question results are in the report. Our guide to approvals across tools covers what to do about all this in the meantime.